Recently in Information Architecture Category
Comparing relational vs. object-oriented database use in content management is highly subjective, and can’t be generalized. That would be like saying that the movie is always better than the book it is based on. Take the Harry Potter series. While the books have been phenomenally successful, the movies are doing equally well at the box office.
There’s not a one-size fits all approach for technology either. Comparing relational and object-oriented databases needs to be done from several perspectives – notably business rationale for the end user as well as technological advantage – not just one.
On the business end, documentation is mission critical, and must be available 24x7. Relational databases like Oracle support application clustering and high availability out of the box. Customers can count on Oracle always being available, and in a global working environment, everyone can get their job done.
Many businesses need to migrate from some form of binary documentation to XML, but it doesn't happen instantly. Using a relational database, these businesses can store their binary documentation and take full advantage of a CMS while they undertake the process of converting to XML. A relational database can also act as a single repository that stores both XML and binary content, eliminating the need for a separate file system and creates a more homogenous environment for IT.
When business demands and technology realities meet, an argument can be made that a mission critical database application like Oracle requires an amount of care and feeding to be properly maintained. There is, however, also a misunderstanding that with an XML database, an end user can simply let it run and everything is fine.
In reality, many companies like to have control over their “family jewels,” and may want the option of feeding other applications that have canned integrations to relational databases. XQuery may be great, but businesses need to search for content that can be in many forms, XML, PDF, Word, etc. Using a relational database and other technologies, it is possible to support a very robust search mechanism across over 295 different formats.
In both cases, scalability is always a concern. The user must be able to scale and manage both vertically (larger machines) and horizontally (additional machines) while maintaining the integrity of the data and 24x7 access to the system.
Relational databases provide out-of-the-box horizontal scalability, as well as the ability to acutely control how system resources are used. This is crucial in serious business applications. Relational databases can stuff entire areas of XML into a single row (such as a
In the native XML database model, the users would end up with hundreds of rows in their database because each tag is stored separately. Even if all the users wanted to do is repurpose a section, they would need to handle every single row.
The proof lies in customer deployments. Many companies have replaced object oriented databases in large part because they didn’t scale. Consequently they’ve been able to grow into very large solutions using a relational database. In fact, one global customer expects to manage a terabyte of data in their (Contenta) CMS by year’s end. Now that’s scalable.
Just as there are many business and technology needs, there are many viable alternatives, including relational and object-oriented databases. To dismiss an entire technology because of one company’s recent acquisition is a blatant sales pitch at best, and technological ignorance at its worst.
I’m not at the conference this week (we do have several people from our shop there), but to answer a question posed by Frank Gilbane:
What is the future of software as a service, and is it appropriate for enterprise content applications like content management, authoring, etc.?
The SaaS model seems to have been proven to the point where it’s hard to imagine that it won’t keep growing. We recently posted a whitepaper on SaaS myths, which debunks most of the common arguments against SaaS. In addition to that discussion, I’d offer the following four points:
1) SaaS is a proven technology. It arguably extends to the early days of the web with software ASPs. I guess you could even argue the lineage goes all the way back to mainframe apps! :-) Certainly, though, the existing SaaS companies have been working successfully with this business model for more than six years now.
2) Web technologies have reached a point where SaaS is an out of the box solution. You can now count on fast network connections for users both in the office and home. Security systems are complete from SSL with web browsers up through terminal services like Citrix which allow even HIPAA compliance. For web apps, browser technologies like IFRAMES and AJAX allow apps to be easily integrated on a page (mashups).
3) SaaS provides much more robust server management and security, especially for small and medium sized businesses. As web applications grow more complex, SaaS allows much more convenient, rigorous and cost-effective control over hosting. By centralizing and focusing, the best resources can be brought to bear on fewer hosting environments.
4) This one is a bit of a prediction, and is specific to web sites. Currently, you have a couple options when adding components like blogs, rss, ecommerce, polls, surveys, and search to web sites. You can install apps for those services, which allows you to control ad placement and design. The alternative is to use free hosted apps where the ad revenue goes to the SaaS company. So, the logical next step is for high quality hosted apps where the ad revenue is shared with the web site. This is already appearing with sites like MetaCafe. For an advanced CMS, though, I am not sure this will happen since the CMS tends to be the hub for all the other web apps, but it is certainly possible for a basic CMS.
Now available! "Beyond Search: What to do When Your Enterprise Search System Doesn't Work, by Stephen Arnold
